A Hybrid Approach of Automatic Post-Editing for Machine Translation

摘要：

This paper presents a hybrid approach of Automatic Post-Editing (APE) for Machine Translation. The basic idea of our method is to combine Decision Tree Learning algorithm and ? linear interpolation of N-gram language models and Class based N-gram language models. Our purpose is to improve quality of machine translation. In our study, we just focused on the problem of the Japanese particle の(no), which kept a highly complex and representative characteristic one of Japanese particles. We have evaluated the efficiency of our approach in simulation experiments of APE of の (no). Experimental results show that our proposed method offers good performance.

关键词： Hybrid approach machine traslation stslistical language model automatic post-editing

作者: Jin anXu

作者单位: School of Computer and Information Technology, Beijing Jiaotong University, Beijing, China

会议类型: 国际会议

会议名称: 2010 International Conference on Information Security and Artificial Intelligence(2010年信息安全与人工智能国际会议 ISAI 2010)

会议地点: 成都

会议语种:英文

页码: 865-868

在线出版日期: 2010-12-17（万方平台首次上网日期，不代表论文的发表时间）